Energy Audit of water networks

- 1 -
ENERGY AUDIT OF WATER NETWORKS
Enrique Cabrera (member ASCE)
1
, Miguel A. Pardo
2
, Ricardo Cobacho
3
and
Enrique Cabrera Jr
4
.

1
Professor, ITA, Universidad Politcnica de Valencia, Camino de Vera s/n. 46022, Valencia,
Spain. Email: ecabrera@ita.upv.es
2
MSc, PhD student, ITA, Universidad Politcnica de Valencia, Camino de Vera s/n. 46022,
Valencia, Spain. Email: miparpi@ita.upv.es.
3
Assistant Professor, ITA, Universidad Politcnica de Valencia, Camino de Vera s/n. 46022,
Valencia, Spain. Email: rcobacho@ita.upv.es
3
Associate Professor, ITA, Universidad Politcnica de Valencia, Camino de Vera s/n. 46022,
Valencia, Spain. Email: qcabrera@ita.upv.es

ABSTRACT

This paper presents the energy audit of a water network, which is obtained from the energy
equation in integral form, and its time integration extended over a given period (day, month or
year). The analysis allows accounting for all the energy in the system, showing that the energy
balance is maintained. This balance allows can be used to obtain performance indicators to
assess the system from the energetic point of view. From these indicators, it is possible to
identify the improvement actions that will make the system more efficient. This energy audit
requires a previous water balance and the mathematical model of the network, both of which are
necessary to know the energy flows through the systems boundaries.

Keywords: water, energy, audit, efficiency.

INTRODUCTION
Until recently, energy savings in the water industry were an operational issue, a pump efficiency
matter, a process improvement or simply a management efficiency target (Lingireddy and
Wood, 1998). However, due to the periodic energy crisis and the need to reduce greenhouse
emissions, there is an increasing motivation to minimize energy requirements in sustainable
water use (USDE, 2006). In any case, it is understandable that the degree of concern with a
water utility in terms of energy efficiency will depend on the circumstances of the system. If
water is obtained from a desalination plant, the required energy before coming into the system is
at least 3.5 kWh/m
3
(NRC, 2008), and the sensitivity to energy-related issues will thus be high.
However, if the water source is natural and has high quality and no treatment costs, energy
issues will become secondary.
Energy Audit of water networks
- 2 -

The present work will analyze in detail the distribution phase in water networks. Until now the
analyses performed have consisted of dividing the energy paid (kWh), called hereafter shaft
energy, by the volume of water delivered to users (m
3
). This ratio provides for this phase a
global estimation of the energy costs per volume, a value that in California ranges from 0.18 to
0.32 kWh/m
3
(CEC, 2005). The example provided in this paper will show that such a range,
pending more accurate data, is a valid reference. However, it is a global indicator and does not
provide information about how that energy is used along the distribution process, which is the
final objective of the energy audit presented here.

Pelli and Hitz (2000) establish the water-energy relationship in a system in an integral manner
and propose two indicators: the infrastructure indicator and the quality indicator. This is an
interesting practical approach that combines elements external to the network (such as the
efficiency of pumps) but ignores the energy dissipated in friction losses (which depends not
only on the consumed flow rate but also on the leakage level). This issue has already been
considered by other authors, such as Colombo and Karney (2002), although their analysis does
not include the whole network and, when it does, the analysis is particularized for a specific
system (Colombo and Karney, 2005).

This paper originates in the study of the energy equation and presents a complete audit of the
distribution system contained within a control volume (which may be either the full network or
a district metering area). In order to apply Reynoldss transport equation to energy, the different
flow terms at all boundaries need to be known. Therefore, a water audit and a calibrated model
are required. In other words, the hydraulic problem must first be solved. Since water is
incompressible, the mechanical and thermal equations are not coupled (White, 1974), and the
energy problem can be solved after the hydraulic one. It must also be noted that the selected
control volume will not include pumps. The hydraulic power that the pumps deliver is an
external contribution, and their efficiency thus needs to be evaluated independently from the
energy audit presented here. This is not the case of regulation valves, which, where present,
should be considered part of the system, as they are within the control volume boundaries and
influence the problem by increasing friction losses.

The most relevant novelty of the energy balance presented here is the assessment of the final
uses of the energy injected to the system and, specifically, to the energy loss associated with
leakage. Such an energy loss results from two different terms: one associated with the water
leaking out of the network and another one related to the energy dissipated in friction losses due
to the additional flow rate needed to compensate for the leakage while meeting demands.
Performance indicators are later used to characterize the whole energy balance, allowing the
assessment of the energy efficiency of the network as well as the influence of the energy losses
in the optimum pipe renovation period (through a cost-benefit analysis).

The global balances presented to date (Todini, 2000) do not explicitly assess the energy losses
associated with leakage. In the current climate change scenario, the need to clearly establish the
water-energy relationship justifies this work.
Energy Audit of water networks
- 3 -

GENERAL APPROACH TO THE PROBLEM
The integral energy equation can be applied to a control volume with known amounts of water
and energy flowing through its boundaries. This implies solving the water balance and the
mathematical model of a system contained within a control volume defined at will. Its
boundaries define which elements are external (contributing to the energy flow) or internal
(storing or dissipating energy). To illustrate the difference between internal and external
elements, the example here presented includes both types of elements, an external reservoir and
an internal regulation tank. The pumps are external elements, providing shaft work, while
service connections and leaks represent the exit control surface of the system. The mathematical
model is necessary to quantify the energy flows (inherent to water flows) through the
boundaries. Figure 1 represents the control volume and the incoming and outgoing flows of
energy.

Compensation
tank
Control Volume
Reservoir
Pump
Energy through leaks Friction energy Energy suplied to users
Compensation
tank
Control Volume
Reservoir
Pump
Energy through leaks Friction energy Energy suplied to users

FIGURE 1. WATER NETWORK AS A CONTROL VOLUME WITH THE TERMS OF THE ENERGY BALANCE

By applying the energy equation (which is in fact a power balance) in its most general form to
the control volume, Equation (1) is obtained:

( )
} }
+
} } }

c
c
=
c
c
+
c
c
=
CS CV
A d v e d e
t t
W
t
Q
dt
dE

, (1)

where dt dE represents the energy exchange per time unit (sum of the heat exchange,
t
Q
c
c
, and
the work,
t
W
c
c
), e represents the total energy per mass unit and represents the fluid density.
Developing the unit energy term, Equation (2), which is better suited to our application, can be
obtained:

( )
} }

|
|
.
|

\
|
+ + + +
} } }

|
|
.
|

\
|
+ +
c
c
=
c
c
+
c
c
CS CV
shaft
A d v
v
u gz
P
d
v
u gz
t t
W
t
Q

2 2
2 2
(2)
Energy Audit of water networks
- 4 -

Equation 2 can be simplified, given the following assumptions:

1. Water is incompressible and thus constant.
2. The heat flow through the boundaries is zero ( t Q c c =0), a reasonable hypothesis for this
application. In fact, this is a very complex problem under research (Burch and
Christensen, 2007) but it does not affect the final result of the audit. With no heat
transfer, friction will slightly increase the temperature of the flowing water.
3. The shaft work,
shaft
W , is supplied by pumps ( t W
shaft
c c =
P
n
i
Pi Pi
H Q ).
4. The kinetic term ( 2
2
v ) is neglected as in most network analyses.
5. The performed energy analysis is quasi-static with an extended period integration. In
each calculation interval, the energy inside the control volume is constant as far as pipes
are concerned, while the energy stored by internal tanks can change.
6. The flow is uniform at the boundaries.

Under these conditions, the energy equation becomes Equation (3):

+ +
(

+ + +
} } }

c
c
=
Inlets
Ii Ii
Ii
Ii
Outlets
Oi Oi
Oi
Oi CV
n
i
Pi Pi
u gz
P
Q u gz
P
Q zd
t
g H Q
P

(3)

Typical boundary elements of water networks are reservoirs, tanks and pumps (Rossman, 2000).
The energy contribution of the reservoirs, which are external to the system, depends on their
head. From the audits point of view, it is a natural energy compared to the artificial energy
provided by pumps (shaft work). A tanks contribution is also natural, although its elevation is
variable with time. As for the energy flows exiting through the nodes, the energy is reflected in
the piezometric head. For a generic outlet, the (demand) node (i) is
Oi Oi Oi
z P H + = . This value is
dependent on the systems reference for elevations. The location of the origin influences the
final value of the energy indicators. It is advisable to use as the origin of the reference system,
z=0, the lowest node of the system.

From the previous considerations, for a system fed from
N
n reservoirs (constant head)
supplying n network outlet nodes, with
P
n pumping stations providing energy to the system,
and with
C
n compensation tanks, Equation (4) is obtained:

} } }

c
c
+
(

+ = + =
VC
n
i
Ni Ni
n
i
Oi Oi
n
i
Oi Oi
n
i
Pi Pi
n
i
Ni Ni
zd
t
u Q u Q H Q H Q H Q P
N P N

0
(4)

Energy Audit of water networks
- 5 -

The preceding power balance, which refers to the whole system, is similar to the energy balance
in Bernoullis equation, as applied to the ends of a pipe:

The natural power supplied provided by reservoirs and tanks is
N
P =
N
n
i
Ni Ni
H Q .
The artificial power provided by pumps is
P
P =
P
n
i
Pi Pi
H Q .
The total power supplied,
0
P , comprises both.
The term
n
i
Oi Oi
H Q is the exit power through the network outlet nodes, the sum of the
power delivered to users (useful power
U
P ) and the power losses resulting from leakage
(
L
P ).
The term
(

n
i
n
i
Ni Ni Oi Oi
N
u Q u Q is the variation of the internal energy of water with time.
With no heat exchange, the increase of internal energy with time is equal to friction
losses (
F
P ).
The term
} } }

c
c
CV
zd
t
=
Ci
n
i
P
C
=
C
P is the change with time (negative or positive) of the
potential energy in the
C
n tanks belonging the system.

Equation (4) can be expressed in a more compact manner as Equation (5):
.
C F L U P N
P P P P P P + + = + (5)
This equation states that the power supplied to the network is equal to the power delivered to the
users plus the power losses (leakage and mechanical friction), a balance finally adjusted by the
compensation term, provided by the tank inside the system. This compensation term becomes
less relevant as the integration period increases. The new audit also allows separating the
internal power (Todini, 2000) in its two terms (P
L
and P
F
) while also considering P
C
.

Energy Audit of water networks
- 6 -

EXTENDED PERIOD INTEGRATION OF THE ENERGY EQUATION
Equation (5) can be integrated through time in a process analogous to the one followed to
simulate the hydraulic behavior of a water network with time. The integration converts the
power terms into energy terms. The adopted notation is shown in Figure 2.

Node i Node i+1
) ( ) (
k li k ui
t q t q + ) ( ) (
1 1 k li k ui
t q t q
+ +
+
) (
k i
t H ) (
1 k i
t H
+
) ( ) ( ) (
k lj k uj k j
t q t q t q + =
Node i Node i+1
) ( ) (
k li k ui
t q t q + ) ( ) (
1 1 k li k ui
t q t q
+ +
+
) (
k i
t H ) (
1 k i
t H
+
) ( ) ( ) (
k lj k uj k j
t q t q t q + =

FIGURE 2. NOTATION USED TO INTEGRATE THE ENERGY EQUATION

) (
k ui
t q , ) (
k li
t q supplied and leakage flow rate delivered in node i at time
k
t .
) (
k uj
t q , ) (
k lj
t q supplied and leakage flow rate circulating in line j at time
k
t .
) (
k i
t H , piezometric head in node i at time
k
t .

The energies resulting from the integration of Equation (5) for a simulation period totaling time
p
t are listed in Table 1.

TABLE 1. ENERGIES INVOLVED IN THE AUDIT
Energy
Notation
Natural energy (supplied by external sources)
) (
p N
t E
Shaft energy (supplied by pumps)
) (
p P
t E
Useful energy delivered to users ) (
p U
t E
Leakage energy losses
) (
p L
t E
Friction energy losses
) (
p F
t E
Compensation energy (associated with internal system tanks)
) (
p C
t E

The difference between natural and shaft energy makes plenty of sense. When water comes into
the distribution system, it has its own previously acquired energy footprint (take for instance the
transport and treatment energy consumption in kWh/m
3
). In the distribution phase, the natural
energy does not modify this footprint, while the shaft energy is included as a new term.

Energy Audit of water networks
- 7 -
Input energy supplied by the reservoir
The external energy, supplied by reservoirs or external tanks, is
t t H t Q t E
N p n i
i
t k
t k
k Ni k Ni p N
A
|
|
.
|

\
|
=
=
=
=
= 1
1
) ( ) ( ) ( , where is the specific weight of water, ) (
k N
t Q is the flow
rate supplied at the time
k
t , ) (
k N
t H is the piezometric head at time
k
t , and t A is the considered
time interval of integration. In order to perform the analysis in an extended period, it is
necessary to add the different k intervals included in that period ( t k t
p
A = ).

If the head of the external sources is constant, the same equation can be simplified resulting
in =
=
=
N
n i
i
Ni Ni p N
H t E
1
) ( , with
Ni
being the volume supplied by reservoir i during the whole
analysis period and
Ni
H the piezometric head of water in that reservoir. This equation reflects
the incoming natural energy into the system.

Energy supplied by pumping stations (shaft work)
The shaft work supplied by pumps is t t H t Q t E
P p n i
i
t k
t k
k Pi k Pi p P
A
|
|
.
|

\
|
=
=
=
=
= 1
1
) ( ) ( ) ( , where ) (
k Pi
t Q is the
flow rate pumped by station i at time
k
t and ) (
k Pi
t H is the head of the pump. This calculation
needs to be done for all
P
n pumping stations that supply shaft work at the different time
instants k .

Energy supplied to users
The useful energy delivered to the customers is A
|
|
.
|

\
|
=
=
=
=
=
n i
i
t k
t k
k i k ui p U
t t H t q t E
p
1
1
) ( ) ( ) ( , where the
number of demand nodes is n.

Outgoing energy through leaks
Leaks in mathematical models are concentrated in demand nodes and behave as a pressure-
driven demand. Although in the audit they appear as lost energy, they are modeled as energy
leaving the system, which is formally analogous to the energy delivered to users:
A
|
|
.
|

\
|
=
=
=
=
=
n i
i
t k
t k
k i k li p L
t t H t q t E
p
1
1
) ( ) ( ) ( .
Energy Audit of water networks
- 8 -

Friction energy
The energy dissipated in the system due to friction is
( ) A
|
|
.
|

\
|
A + =
=
=
=
=
L p n j
j
t k
t k
k j k lj k uj p F
t t h t q t q t E
1
1
) ( ) ( ) ( ) ( where
L
n is the number of lines of the network and
) (
k j
t h A the friction losses in line j at time
k
t (known from the mathematical model of the
network).

The flow rate in line j , Figure 2, is ) ( ) ( ) (
k lj k uj k j
t q t q t q + = . This equation shows that the leaked
flow rate (flows through the network before leaking out) generates additional friction losses.
The energy audit will determine the value of this additional energy by simulating the networks
behavior with and without leaks. Representing a global balance for the whole water distribution
network and not divided in lines, the individual values, ) (
k uj
t q and ) (
k lj
t q remain unknown,
although for this analysis, this fact bears little relevance.

Energy compensation of the downstream tank
Many networks have a compensation tank to accumulate water during low consumption hours
while releasing it in peak hours. These tanks belong to the system. However, the net flow of
water and energy in one of these tanks, when integrated through a long enough period, is zero,
and so is their contribution to the long-term analysis. During normal operation, with shorter
periods, the tanks can be considered mass and energy sources and sinks and must be included in
the audit.

The variation of potential energy stored in a tank of constant section for a given period of time
is ( ) ( ) ( ) = = A
=
=
=
=
C C
i
n i
i
i p i i
n i
i
C p i C p C
t z t z A t E t E t E
1
1
2 2
1
1
2 / ) ( ) ( ) ( ) ( ) ( , with
i
A being the section of
compensation tank i and ) (
p i
t z , ) (
1
t z
i
the levels of the free surface of water of tank i at the
initial and final times. The maximum variation of this energy,
max C
E A , obviously corresponds to
an oscillation between a full and an empty tank.

GLOBAL ENERGY BALANCE
The energy balance for a given period is the result of integrating Equation (4) over time (energy
results from integrating power with time). These energies (except the compensation term) enter
or exit the system or, otherwise, are dissipated. This enables the derivation of Equation (6):

= + = ) ( ) ( ) (
p P p N p Input
t E t E t E

) ( ) ( ) ( ) ( ) ( ) ( ) (
p on Compensati p Dissipated p Output p C p F p L p U
t E t E t E t E t E t E t E A + + = A + + + =
.
(6)

Energy Audit of water networks
- 9 -

Developing the terms results in Equation (7).

t t H t Q
N p n i
i
t k
t k
k Ni k Ni
A
|
|
.
|

\
|

=
=
=
= 1
1
) ( ) ( + t t H t Q
P
p n i
i
t k
t k
k Pi k Pi
A
|
|
.
|

\
|

=
=
=
= 1
1
) ( ) ( =
= A
|
|
.
|

\
|

=
=
=
=
n i
i
t k
t k
k i k ui
t t H t q
p
1
1
) ( ) ( + A
|
|
.
|

\
|

=
=
=
=
n i
i
t k
t k
k i k li
t t H t q
p
1
1
) ( ) ( +
+ ( ) A
|
|
.
|

\
|
A +
=
=
=
=
L
p n j
j
t k
t k
k j k lj k uj
t t h t q t q
1
1
) ( ) ( ) ( + ( ) ( )
=
=
C
n i
i
i p i i
t z t z A
1
1
2 2
2 / ) ( ) (
.
(7)

The aforementioned compensation term is only relevant in short-term simulations. The
threshold value,
B p
t
,
, separating the short term from the long term is established by imposing that
the maximum compensation energy is only a small percentage of the system energy input
(
Input
E ). For a 1% value, this threshold is:
) (
) (
max ,
,
daily E
E
days t
Input
c
T p
100
1
A
=
.
(8)

Therefore, if the energy audit is assessed annually, the compensation term may be withdrawn,
and Equation (6) becomes Equation (9). Table 2 summarizes a long-term audit:

) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (
p Dissipated p Output p F p L p U p P p N p Input
t E t E t E t E t E t E t E t E + = + + = + =
.
(9)

TABLE 2. ENERGY BALANCE FOR A DRINKING WATER NETWORK ON THE LONG TERM
) (
p Input
t E
) (
p N
t E
(natural input energy)
) (
p U
t E
(energy delivered to users)

) (
p Output
t E
) (
p L
t E
(outgoing energy through leaks)

) (
p P
t E
(shaft input energy)
) (
p F
t E
(friction energy)

) (
p Dissipated
t E

Energy Audit of water networks
- 10 -

BASIC ENERGY INDICATORS
Energy indicators have traditionally been expressed in kWh/m
3
as the ratio between the billed
shaft work and the metered volume. The indicators proposed here provide an overall and
intuitive assessment of the system and have sense mostly in the long term. The piezometric
heads are set to the lowest node, which is taken as the z=0 reference, a criterion that affects the
values of the indicators (as all the analysis is carried out by means of piezometric heads)
deeming impossible the comparison with other systems.

Context information
Each system is, from an energetic point of view, different. The network can be flat or hilly, and
it may or may not require intermediate pumping stations. It is obvious that a system supplied
from a source located at a higher point represents an ideal situation, as it needs no pumping at
all. The opposite case is a system supplied from groundwater where every cubic meter requires
a significant amount of energy before it reaches users. The difference in context between these
two situations is covered by the first context indicator
1
C (formally, it cannot be considered
context information - see the definition in Alegre et al. 2006 - as it may slightly change
depending on leakage or friction, but in our experience, it is a constant enough value and
provides valuable context data to compare utilities).
1
C (Table 3) shows which portion of the
energy delivered to the system is natural and ranges from 0 to 1, with the maximum being
reached when all the injected energy is gravitational, being provided by a high water source.

The second context information item
2
C , Table 3, takes into account how demanding from an
energy point of view the network is. As the ratio between the minimum useful energy
useful
E
min,

defined in each node from the minimum required head ( /
, Mn i i Mn
P z h + = ) and a theoretical
minimum required energy (for a flat, leak free and frictionless network)
flat
E
min,
. Since this ideal
network corresponds to a flat layout with all nodes located at the same maximum height z
max
,
the best possible value of
2
C is one.

TABLE 3. CONTEXT INFORMATION
1
C

Energy nature
2
C
Network energy requirement
) (
) (
p Input
p N
t E
t E
C =
1

( )
) (
) (
) (
) (
min
min, ,
min
min,
min,
p U
n
i
i p i u
p U
t k
t k
n
i
i Min k ui
flat
useful
t
P
h t
t
P
t h t q
E
E
C
p


=

A
= =
=
=
= =

1
1
2
1


Energy Audit of water networks
- 11 -
Efficiency indicators
As defined in the IWA manual of performance indicators (Alegre et al., 2006) the context
information items provided above are useful for characterizing the system. However, they
cannot be changed by management decisions and thus should not be used to measure how well
the system is managed. In order to perform such an analysis, five performance indicators are
proposed (Table 4):

TABLE 4. ENERGY EFFICIENCY INDICATORS
1
I

Excess of supplied energy
2
I

Network energy efficiency

3
I

Energy dissipated through friction

=
=
n
i
i p i u
p input
h t
t E
I
1
1
min, ,
) (
) (
u

) (
) (
p Input
p U
t E
t E
I =
2

) (
) (
p Input
p F
t E
t E
I =
3

4
I
Leakage Energy
5
I

Standards compliance
) (
) ( ) ( ) (
'
p Input
p F p F p L
t E
t E t E t E
I
+
=
4


=
=
n
i
i p i u
p U
h t
t E
I
1
5
min, ,
) (
) (
u


The first indicator,
1
I , is the ratio between the real energy entering the system and the
minimum useful energy.


2
I , is a measure of the efficiency of the use of the energy injected to the system (which
fraction of the total energy input is useful).


3
I

represents the hydraulic capacity of the network. A higher value indicates lower
efficiency. Although this can be brought to values very close to zero, eliminating friction
losses implies a very costly design. Target values depend on a balance between
investment and running costs.

The fourth indicator,
4
I , measures the energy loss due to leakage, which results from the
sum of energy loss through leaked water ) (
p L
t E and the additional energy required to
overcome friction with the increased flow rate needed to overcome leakage (difference
between the actual energy dissipated in friction losses ) (
p F
t E and the value of friction
losses in a leak-free network, ) (
'
p F
t E ).

Finally,
5
I , is the direct ratio between the energy delivered to users and the minimum
required useful energy. It is a network-level indicator that averages the overall condition
of the system but may leave sector performance unnoticed (the average condition may
be good while some sectors are performing poorly). At a first glance,
5
I can be
interpreted as follows:
Energy Audit of water networks
- 12 -

5
I <1 shows that average pressure levels are insufficient and below standards. The
minimum value for a given network appears when water is delivered to users by
means of underground tanks located in users households. The supply pressure
would then be atmospheric and the numerator of
5
I equal to: A
|
|
.
|

\
|

=
=
=
=
n i
i
t k
t k
i k ui
t z t q
p
1
1
) (

5
I >1 is the most common case. The pressure is kept above the service standards. A
value closer to 1 indicates greater efficiency in meeting them. On the other hand,
such an excess of potentially recoverable energy depends mainly on the minimum
excess pressure value p
min
(the minimum value for all nodes and all simulation
intervals of the difference between the real pressure and the minimum allowable
value p
service
p
min
). However, the actual possibility of recovering such energy
greatly depends of the characteristics and operation of the network. Another
possibility is that even when
5
I >1, some node may not satisfy the pressure standard,
in which case p
min
would be negative.

NUMERICAL EXAMPLE
To illustrate both the energy audit and the use of the proposed indicators, a numerical example
is presented. As water balances are usually assessed in coincidence with meter-reading periods
(month, trimester or year) usually the audits will be assessed over the long term.

Basic data
Figure 3 shows the network layout while Table 5 shows the node and line data. Node 33, which
is the lowest point in the network, becomes the origin for elevations and all other values are
given in reference to this elevation. The simulations were carried out using the EPANET 2.0
toolkit.

Reservoir
9
111
11 12 13
21 22 23
31 32 33
10 11 12
1
21 22
31 32
Pump
121
121
122
113
123
Tank
10
Reservoir
9
111
11 11 12 12 13 13
21 21 22 22 23 23
31 31 32 32 33 33
10 11 12
1
21 22
31 32
Pump
121
121
122
113
123
Tank
10 10

FIGURE 3. GENERAL LAYOUT OF THE NETWORK

Energy Audit of water networks
- 13 -
TABLE 5. LINE AND NODE DATA
Line Length (km)
Diameter
(mm)
Node
Base
demand
(l/s)
Elevation
(m)
Emitter
coefficient
(m
3-o
/s)
10 2 400 Node 10 0 5.8 0.002611285
11 2 300 Node 11 5 5.8 0.010445142
12 2 350 Node 12 5 4 0.010445142
21 2 200 Node 13 3 2 0.010445142
22 2 200 Node 21 5 4 0.013056427
31 2 200 Node 22 6.5 2 0.015667712
111 4 200 Node 23 5 0 0.013056427
112 4 250 Node 31 3 4 0.007833856
113 4 300 Node 32 3 5 0.010445142
121 4 200 Node 33 3 0 0.007833856
122 4 200 Reservoir - 25 -
123 4 200 Tank - 32 -
32 2 200
1 2 400

TABLE 6. HOURLY COEFFICIENTS OF WATER DEMAND MODULATION
Time 1 2 3 4 5 6 7 8 9 10 11 12
Coefficient 0.6 0.5 0.45 0.45 0.5 0.5 0.9 1.3 1.4 1.1 1.5 1.4
Time 13 14 15 16 17 18 19 20 21 22 23 24
Coefficient 1.4 1.45 1.45 1.3 1.2 1.2 1.1 1.1 1.2 1.1 0.9 0.7

The pipe roughness is 0.1mm. The diameter of the compensation tank is 20 m, and its level
oscillates between 2.5 m (initial value for the simulation) and 7 m (maximum value). The
minimum node pressure (22 mcw) is maintained by a pump (characteristic
curve:
2
003646 0 33 93 Q H = . . ). The pump starts and stops when the water level reaches the limits
in the tank. The two simulation periods correspond to one day for the short term and one year
for the long term, a value clearly above the long-term threshold ( in our example, as shown
later,
T p
t
,
= 16 days).

The initial annual water audit (apparent losses are included in the delivered water) is:

Injected water: ) (
p N
t = 1.743 hm
3
/year.
Delivered water: ) (
p U
t = 1.214 hm
3
/year.
Real losses: ) (
p L
t = 0.529 hm
3
/year.

Both the total user demand and the hourly modulation coefficients are constant along the year.
Table 6.

Energy Audit of water networks
- 14 -

The leakage flow rates at the nodes are determined supposing that they are proportional both to
the pressure in the node (pressure driven demand) and to the length of the pipes linked to it
(uniform distribution). Additionally, the continuity equation needs to be fulfilled (and therefore
the sum of all volumes leaked through the nodes must be in accordance with the water audit
results). Each nodal leak is characterized through the corresponding emitter, which is adjusted
by successive approximations in a quick convergence method described in Almandoz et al.
(2005). The characteristics of the emitters follow the EPANET model | |
o
) ( ) (
, k i i E k li
t H C t q A =
(Rossman, 2000), where
i E
C
,
(m
3-o
/s) is the coefficient assigned to each node, ) (
k i
t H A (m) the
pressure variation through the leak and o =1.2 the emitter exponent that models the
characteristics of the pipe material. The resulting emitters coefficients are also depicted in
Table 5.

Results
Table 7 shows the results of the four analyzed cases. They correspond to daily and annual
simulations for both an ideal network (no leaks) and a real network. While the first period can
be considered a short one, the second qualifies as a long-term simulation, with a period much
longer than the =
T p
t
,
16 days required by Equation (8), as the input energy is ) (
p input
t E = 1364.41
kWh/day and the maximum variation in the compensation tank is
max , c
E A = 218.62 kWh. The
hydraulic time step used to calculate the simulations is 1 minute for the short-term simulations
and 15 minutes for the long-term ones.

TABLE 7. ENERGY BALANCE (ALL CASES PRESENTED)
Energy

Real network

Ideal network (no leaks)
Short Term
T p p
t t
,
<
(kWh/day)
Long Term
T p p
t t
,
>
(MWh/year)
Short Term
T p p
t t
,
<
(kWh/day)
Long Term
T p p
t t
,
>
(MWh/year)
) (
p Input
t E
) (
p N
t E 386.66 (28.3%) 118.75 (28.7%) 322.52 (27.3%) 82.79 (27.7%)
) (
p P
t E 977.75 (71.7%) 294.76 (71.3%) 857.95 (72.7%) 215.60 (72.3%)
) (
p Output
t E
) (
p U
t E 453.20 (33.2%) 169.13 (40.9%) 501.71 (42.5%) 179.73 (60.2%)
) (
p L
t E 232.67 (17.1%) 77.53 (18.8%) - (0%) - (0%)
) (
p C
t E A 128.87 (9.4%) 0.17 (0.0%) 217.28 (18.4%) 0.20 (0.1%)
) (
p Dissipated
t E ) (
p F
t E 549.84 (40.3%) 166.86 (40.4%) 461.12 (39.1%) 118.63 (39.8%)

The theoretical energies, defined as
useful
E
min,
and
flat
E
min,
, are equal to 219.67 kWh/day
(80.18 MWh/year) and 199.42 kWh/day (72.79 MWh/year), values that correspond to an almost
flat network. Additionally, the results in Table 7 show that:

Energy Audit of water networks
- 15 -
In a leak-free network, there are additional advantages to the obvious energy input
savings (1180.47 kWh/day versus 1364.41 kWh/day). Additionally, the network
performance is clearly improved. This is shown by the amount of energy received by
users (501.71 kWh/day versus 453.20 kWh/day) with all of the improvement
possibilities that come with it. The leak-free system involves a higher value of
5
I ,
revealing to what extent leakage consumed extra capacity already existing in the system,
and that such extra capacity is an opportunity for further improvement in energy
performance by reducing pumping output to curtail unnecessary surplus pressure at the
delivery points. It also reveals a better situation for future network expansions.
Quantifying these improvements delivers the pmin value, which for a leaking network
is 1.05 m.c.w. for the short-term simulation and 0.21 m.c.w. for the long-term one. The
values increase to 6.85 m.c.a. and 6.69 m.c.a., respectively, in the case of a leak-free
network. These minimum values are registered when the network input point is the
compensation tank in node 31. The partial or total recovery of these energy surpluses
requires the optimization of the operating conditions of the network.

The energy loss associated with leaks (which is lost directly through them) is 232.67
kWh/day, but the total energy loss is larger when the additional friction losses are
considered (the difference between the 549.84 kWh/day and the 461.48 kWh/day of the
leak free network). This brings the total daily savings to 321.03 kWh/day (which
represents 27.2% of the total energy in use).

If percentages are taken into account, the audit shows that while there is little variation in
the input energy, there are great differences when considering the energy delivered to
users. The differences observed between the two scenarios are to a great extent due to
the significant participation in the short term of the compensation energy in the tank.

The energy dissipated in friction losses ) (
p F
t E =166.86 MWh/year is equal to the
increment in internal energy of ) (
p F
n i
i
Ni Ni
n i
i
Oi Oi
t E u Q u Q
N
=
|
|
.
|

\
|

=
=
=
= 1 1
provided that no heat is
exchanged. Under these conditions, the water temperature increases by 0.08C on
average (being =1000 kg/m
3
,
p
C =4180 J/kg/K and T C u
p
A = A ).

Energy assessment of the network with the new indicators
The aforementioned indicators defined can help to assess, compare and improve the energy
efficiency of different networks. They can be calculated with the results of both short-term and
long-term energy audits, although short-term analyses may create distortions in the indicators
values.

Utilities often use the shaft energy per volume indicator (kWh/m
3
), which can be referred to for
both injected and consumed volumes. However, when the objective is to assess the energy
efficiency, it makes sense to use the delivered volume as a denominator, for it is the ultimate
Energy Audit of water networks
- 16 -
goal of the utility (i.e., to provide users with a certain volume of water). This can clearly be seen
when comparing both possibilities in our example. For the ideal network, both volumes are the
same, and so are the indicator values (0.17 kWh/m
3
). In the real network, the values are 0.17
kWh/m
3
when referring to the injected volume and 0.24 kWh/m
3
when referring to the delivered
volume. It must thus be underlined that if the energy intensity of injected water is calculated, a
leaking network would appear to be as energy efficient as a leak-free one (0.17 kWh/m
3
in both
cases). When the comparison is carried out taking into account the delivered volume, the
absence of leaks becomes a clear energy advantage (0.17 vs. 0.24 kWh/m
3
).

Utilities become sensitive to this indicator, as their energy sources are becoming more
expensive. Although full gravity-driven systems would not feel the urgency to account for this
indicator, as the energy they use appears 'free', within a broader context of resource scarcity,
higher water and energy prices, and the pressure any future climate change might place on
these, opportunities for recovering existing capacity become more relevant, and the opportunity
cost of energy inefficiency more apparent. The possibility of employing micro-turbines to
recapture energy in cases where chronic dissipation would otherwise be the norm ought to be
entertained when feasible.

Table 8 presents for both scenarios the value of the new energy indicators assessed with the
results of annual audits. All of them improve as leaks disappear. In the case of
5
I this is due to
the increase of useful energy delivered to users. Especially remarkable is the improvement of
4
I ,
although it comes as no surprise, as this indicator is closely related to leakage values.

TABLE 8. ENERGY INDICATORS

1
C
2
C
1
I
2
I
3
I
4
I
5
I
Real network 0.28 1.10 5.16 0.41 0.40 0.30 2.11
Ideal network 0.28 1.10 3.72 0.60 0.40 0.00 2.24

Context information, as expected, is independent of the state of the network (such is the
condition to be considered context, Alegre et al. 2006). The first context information item shows
that less than 30% of the injected energy is natural.
2
C
.
with a value close to 1, indicates that the
network is quite flat.

The first indicator
1
I shows that the input energy of the network is more than 5 times the
minimum amount of energy necessary to supply the service, leaving significant room for
improvement. As a matter of fact, when leakage disappears, this indicator is brought down to
3.72. Additional strategies to improve this indicator could include re-designing the network
diameters to reduce friction losses or installing variable-speed pumps to better adjust to pressure
requirements.

Energy Audit of water networks
- 17 -
2
I shows which percentage of the energy is delivered to the users (41% in the real network).
This leaves 59% of the energy being lost through either leakage or friction. In a leakage-free
scenario, the value reaches 60% (representing an almost 50% improvement).

The third indicator shows how much energy is used to overcome friction in pipes. In this case, a
value this high (40%) indicates that the length of the network is very significant, the diameters
are tight, or a combination of both factors. The value is the same for both scenarios, which is
understandable in a leak-free situation, as the friction energy is reduced, but so is the input
energy. In any case, a value of 40% is high enough to trigger the substitution of key pipes with
larger ones.

Of all the indicators,
4
I

is probably the most innovative one. It shows which fraction of the
energy entering the system is lost due to leakage. A deeper analysis could include energy use
outside the distribution stage, and it could take into account, for instance, whether water comes
from a desalination plant or from a surface source. In other words, such an analysis would
require accounting for the water energy footprints corresponding to all other stages previous to
distribution (from abstraction to bulk transport).

Finally,
5
I shows the excess of energy delivered to users, that is, the surplus of energy delivered
on top of the minimum energy required to meet the pressure standards. The value of this
indicator can be improved by means of regulation elements such as valves or variable-speed
pumps.

This performance assessment system is a useful tool to determine the optimal strategies to
improve the energy management of the system. The indicators are also useful for assessing the
potential for improvement.
Energy Audit of water networks
- 18 -

CONCLUSIONS
Until a decade ago, the most relevant aspect of the water-energy relationship was hydroelectric
production. Today, the focus has shifted towards water as an energy-consuming agent. This new
perspective has turned water distribution into a very relevant stage and has brought attention not
only on how much energy is consumed but also on how that energy is used. A proper
assessment of how much energy is wasted as a result of network leakage is the main aim of this
paper. Such energy loss results not only from the energy leaving the system through leaks
(which can be quite significant depending on the energy footprint of the produced water, e.g.,
desalinated water) but also the energy needed to overcome additional friction losses created by
higher circulating flow rates through the pipes.

The audit presented in this work can be used to identify the end uses of the energy entering the
network and thus to define a performance assessment system that characterizes the network
from an energy perspective through context information items and evaluates its energetic
performance. The energy audit approach can also be supplemented with water and energy price
information, as well as estimates of carbon and GHG impacts for the sources and amounts of
energy use, in order to form part of a more holistic evaluation of system performance
improvement options, such as might be undertaken in a cost-benefit analysis framework. As a
matter of fact, these tools could easily be used from a regulatory or administrative perspective to
create incentives for a more efficient use of energy in water distribution. The energy audit, like
associated indicators, requires a previous water audit and that both of them are applied in similar
conditions (either to the whole network or a sector). The energy audit can be extended to any
period of time, but as the water balance is usually available for a year, the audit should cover the
same period.

The energy audit is carried out with a simple but lengthy calculation process, manipulating
EPANET result files. The software is not able by itself to calculate the suggested indicators.
Should EPANET incorporate this option, the energy audit of the system would become an
instant tool available for all technical network managers.

ACKNOWLEDGEMENTS
The authors would like to acknowledge the valuable contributions of the reviewers of this paper,
as their comments and suggestions have helped to significantly improve its contents. The
authors would also like to thank the Spanish Ministry of Science and Innovation for its support
through the research project CGL2008-01910/BTE.

Energy Audit of water networks
- 19 -

NOTATION
i
A Section of the i tank
i E
C
,
Emitter coefficient at node i.
p
C Specific heat of water
e Energy per mass unit
) (
p C
t E Energy compensation of the tanks for the simulation period
) (
1
t E
i
C
Energy compensation of the tanks at the initial time of a simulation
) (
p C
t E
i
Energy compensation of the tanks at the final time of a simulation
) (
p F
t E Friction energy for the simulation period
) (
'
p F
t E Friction energy in a leak-free network
) (
p Dissipated
t E Energy losses due to friction for the simulation period
) (
p Input
t E Input energy for the simulation period
) (
p L
t E Energy through leaks for the simulation period
useful
E
min,
Minimum useful energy needed in a frictionless, leak-free network served with the minimum required
pressure.
flat
E
min,
Minimum theoretical energy needed in an ideal network, frictionless, leak-free and flat.
) (
p N
t E Energy supplied by the reservoirs for the simulation period
) (
p Output
t E Output energy for the simulation period
) (
p P
t E Energy supplied by pumping stations for the simulation period
) (
p U
t E Energy supplied to users for the simulation period
) (
k i
t H Piezometric head at node i at time interval
k
t
i Min
h ) ( Minimum required piezometric head at node i
) (
k Ni
t H Piezometric head at the reservoir i at time interval
k
t
) (
k Pi
t H Piezometric head of the I pump at time interval
k
t
Ni
H Piezometric head of water at reservoir, ) (
k Ni Ni
t H H =
1
C Energy nature coming into the system
2
C Energy management complexity of the network (context information)
1
I Excess of supplied energy (performance indicator)
2
I Network energy efficiency (performance indicator)
3
I Energy dissipated through friction (performance indicator)
4
I - Leakage energy (performance indicator)
5
I Standards compliance (performance indicator)
n Number of demand nodes of the network
C
n Number of compensation tanks of the network
L
n Number of pipes of the network
N
n Number of reservoirs
P
n Number of pumps
C
P Power supplied by compensation tanks
Energy Audit of water networks
- 20 -
F
P Power required to overcome friction losses
L
P Power lost through leaks
N
P Power provided by reservoirs
0
P Total power supplied
P
P Power provided by pumps
U
P Power delivered to users

) (
k i
t P
Pressure at node i at time interval
k
t

Min
P
Minimum pressure required by standards at any node and any time
) (
k j
t q Flow rate at line j at time interval
k
t
) (
k li
t q Leakage flow rate at node i at time interval
k
t
) (
k lj
t q Leakage flow rate at line j at time interval
k
t
) (
k ui
t q Consumed flow rate at node i at time interval
k
t
) (
k uj
t q Consumed flow rate at line j at time interval
k
t
Q Heat exchange at the control volume
) (
k Ni
t Q Flow rate supplied by reservoir i at time interval
k
t
) (
k Pi
t Q Flow rate supplied by pumping station i at time interval
k
t
k
t Time interval of the steady state simulation
p
t Total time of simulation (long or short)
T p
t
,

Short-term/ long-term time threshold (days)

u Internal energy per mass unit
) (
p L
t Total leakage volume for the simulation period
) (
p N
t Total volume injected for the simulation period
) (
p Output
t Consumed volume for the simulation period
) (
p U
t Total volume consumed by users for the simulation period
) t (
p i , u
u Total demand of node i during the simulation period
p
t
W Work Exchange at the control volume
i
z Elevation of node i
) (
1
t z
i
Water level in the compensation tank i at the initial time of a simulation
) t ( z
p i
Water level in the compensation tank i at the final time of a simulation
o Emitter exponent
Specific weight of water
max C
E A Maximum compensation energy of the tank
) (
k j
t h A Friction losses in line j at time interval
k
t
min
p A

Excess pressure minimum value (for all nodes and simulations) calculated as the minimum difference
between the real pressure and the minimum required pressure.
t A Time interval of integration (
k k
t t t = A
+1
)
) (
p T
t A Volume stored in the compensation tank for the simulation period
Density of water

Energy Audit of water networks
- 21 -
REFERENCES
Alegre H., Baptista J.M., Cabrera E. Jr., Cubillo F., Duarte P., Hirner H., Merkel W., Parena R.,
(2006). Performance Indicators for Water Suplly Services, IWA Publishing.

Almandoz, J.; Cabrera E., Arregui, F., Cabrera Jr. E. and Cobacho R. (2005). Leakage
Assessment through Water Distribution Network Simulation. Journal of Water resources
Planning and Management, 131(6), 458-466.

Burch, J., and Christensen, C. (2007) Towards Development of an Algorithm for Mains Water
Temperature. Proceedings of the 2007 ASES Annual Conference, Cleveland, OH.

CEC (California Energy Commission), (2005). Californias Water-Energy Relationship
report. California

Colombo A.F. and Karney B.W. (2002). Energy cost of leaky pipes: Toward comprehensive
picture. Journal of water resources, planning and management, 128(6), 441-450.

Colombo A.F. and Karney B.W. (2005). Impacts of leaks on Energy Consumption in Pumped
Systems with storage. Journal of water resources, planning and management, 131(2). 146-
155.

Lingireddy, S. and Wood, J. (1998) Improved operation of water distribution systems using
variable-speed pumps. Journal of Energy Engineering, 124 (3), 90-103.

NRC (National Research Council), 2008 Desalination a national perspective. NAP Press,
Washington, D.C.

Pelli,T. and Hitz, H.U. (2000). Energy Indicators and savings in water supply. Journal
American Water Works Association, Vol92 (6), June 2000.55-62.

Rossman, L. A. (2000). EPANET 2: Users manual, U.S. Environmental Protection Agency,
Cincinnati.

Todini E., 2000 Looped water distribution networks design using a resilience index based
heuristic apprach Urban Water 2 (2000) pp 115-122.

USDE (United States Department of Energy), (2006). Energy Demands on Water Resources.
Report to congress on the interpendency of energy and Water.

White, F.M. (1974) Viscous fluid flow. B.J. Clark and Michael Gardner eds, McGraw-Hill,
New York. Chapter 3. Solutions of the Newtonian viscous-flow equations.